Systems with multiple transmitting antennas and multiple receiving antennas (MIMO: Multiple Input Multiple Output) for transmitting multiple pieces of data between different points have been known [B1]. Since the channel capacity of the MIMO system increases linearly with the number of transmitting antennas thereof [B2], such systems have drawn quite a lot of attention. In order to achieve the highest throughput, each transmitting antenna transmits an independent stream of data that has been generated by serial-to-parallel conversion of the data that is to be transmitted.
In practice, two modes of implementation of such systems are available. Of these modes, a system according the first mode transmits the signal after modulating the signal using OFDM (orthogonal frequency division multiplexing) technique [B3], ([B4]. A system according to the second mode transmits the signal without any OFDM being done [B5]. In a channel, the signals from different transmitting antennas are mixed up in space, while, at the receiver, signal processing technique for multiple receiving antennas is used to separate the signals. These two system of modes simultaneously transmitting multiple signals can be represented using mathematical models shown in Eq.(1) and Eq.(2), respectively.
For an OFDM-MIMO system, a received signal on each of the frequency bins can be expressed as:uk=Hxk+nkεCMx1  (1),where
  H  =            [                                                  h              11                                                          h              12                                            …                                              h                              1                ⁢                                                                  ⁢                N                                                                                        h              21                                                          h              22                                            …                                              h                              2                ⁢                N                                                                          ⋮                                ⋮                                ⋱                                ⋮                                                              h                              M                ⁢                                                                  ⁢                1                                                                        h                              M                ⁢                                                                  ⁢                2                                                          …                                              h              MN                                          ]        .  The OFDM-MIMO system is a system which uses no convolution operation.
On the other hand, for a non-OFDM system with a channel that is frequency selective [B5], the received signal can be expressed asuNo=HNo*xk+nkεCMx1  (2),where the superscript of “*” indicates convolution operation, and
                              H          No                =                              [                                                                                                      ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                          11                                                ⁡                                                  (                          k                          )                                                                                                                                                                                ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                          12                                                ⁡                                                  (                          k                          )                                                                                                                                      …                                                                                            ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                                                      1                            ⁢                                                                                                                  ⁢                            N                                                                          ⁡                                                  (                          k                          )                                                                                                                                                                                                            ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                          21                                                ⁡                                                  (                          k                          )                                                                                                                                                                                ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                          22                                                ⁡                                                  (                          k                          )                                                                                                                                      ⋯                                                                                            ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                                                      2                            ⁢                                                                                                                  ⁢                            N                                                                          ⁡                                                  (                          k                          )                                                                                                                                                                  ⋮                                                  ⋮                                                  ⋱                                                  ⋮                                                                                                                        ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                                                      M                            ⁢                                                                                                                  ⁢                            1                                                                          ⁡                                                  (                          k                          )                                                                                                                                                                                ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                                                      M                            ⁢                                                                                                                  ⁢                            2                                                                          ⁡                                                  (                          k                          )                                                                                                                                      …                                                                                            ∑                      k                                        ⁢                                                                  δ                        ⁡                                                  (                          k                          )                                                                    ⁢                                                                        h                          MN                                                ⁡                                                  (                          k                          )                                                                                                                                          ]                    .                                    (        3        )            Equations (2) and (3) implies that a number of filters must be used for estimating an inverse transfer characteristic of a MIMO channel. In order to obtain the transmitted signal xk from the received signal uNo, it is necessary to use HNo at the receiver as described above. Such a kind of operation is computationally expensive in comparison to the signal separation in the OFDM-MIMO system. In a non-OFDM system, the receiver has to estimate inverse filters to estimate the inverse characteristic of the channel and this causes increasing in the complexity of the receiver. However, in the OFDM-MIMO system, there is no convolution operation and it is not necessary to estimate the inverse filters.
In the related art, channel state information or inverse channel must be estimated before signals are separated. Such estimation is achieved by sending a training signal, which is used at the receiver to estimate the channel. However, it is necessary to pay an expense of a reduced throughput due to the transmission of the training signal. An alternative approach is a complete blind identification of MIMO CSI (Channel State information) or ICSI (Inverse Channel State information) [B1], [B2], [B6]-[B8]. With this, however, there would be the possibility of separated sources producing permutation of the originally transmitted signals, that is, permutation of the order [B8]. Such permutation would have to be removed by using a maximum likelihood detector (MLD), which considers all possible column-direction combinations and selects the combination with minimum error.
As a solution to the problem of occurrence of permutation in the blindly-separated sources, a tagging scheme based on assigning unique filter to each transmitter was proposed [B9]. The unique filter is a filter having different characteristics for each transmitter. With this technique, each constant modulus source signal is filtered by a unique filter before transmission. At the receiver, an inverse filter is used to convert the non-constant modulus signal back to a constant modulus signal. However, since each source has a unique filter, only the source of interest will be retrieved, and, as a result, the separated signals will be free of source permutations.
However, in such a tagging scheme, the resulting signal may experience a phase rotation as in any other blind algorithm for a single input single output (SISO) system. A solution to this problem is the differential QPSK mode of modulation [B10]. It is also possible to use a channel decoder, which is simpler in configuration since it only involves dealing with a single source. Now, since each transmitting antenna is assigned a unique filter, the length of these filters must also be increased when the number of transmitting antennas increases so as to maintain the uniqueness of the filters. Such an increase in the filter length will increase the computational complexity of the receiver [B9]. When an allpass is used, the latency of the system will also increase.
FIG. 1 shows a baseband model of an OFDM-MIMO system based on the tagging filters. For explanations purpose, a constant modulus signal is assumed. This system includes, on the transmission side, a plurality of unique filters 101 and antennas 102 provided for respective unique filters. Here, N (N≧2) pieces of unique filters 101 are provided. The signals from transmitting antennas 102 reach the reception side having M pieces of receiving antennas 103. At the reception side, further arranged are filters 104 having inverse characteristics of respective filters 101 on the transmission side, and signal processors 105 provided for respective filters 104. Signal processors 105 perform blind MIMO signal processing. On the reception side, a signal subjected to blind signal separation is outputted from each signal processor 105.
On the transmission side, after carrying out serial-to-parallel conversion for the data to be transmitted, each element of the parallel data is filtered with unique filter 101 that has an impulse response fTx(l)(n). Here, “(l)” in the superscript corresponds to the index of the buffered input xl(k), where, l=1, . . . , N.
The baseband signal which is transmitted using the l-th antenna is given by:
                                                        y              l                        ⁡                          (              k              )                                =                                    ∑                              n                =                0                            ∞                        ⁢                                                            x                  l                                ⁡                                  (                  n                  )                                            ⁢                                                f                  Tx                                      (                    l                    )                                                  ⁡                                  (                                      k                    -                    n                                    )                                                                    ,                            (        4        )            where |yl(k)|, which is the absolute of yl(k), is not a constant for all values of k. For a flat MIMO channel, the baseband signal received by at antenna 103 with antenna number j will be given by
                                                        v              j                        ⁡                          (              k              )                                =                                                    ∑                                  l                  =                  1                                N                            ⁢                                                h                  jl                                ⁢                                                      ∑                                          n                      =                      0                                        ∞                                    ⁢                                                                                    x                        l                                            ⁡                                              (                        n                        )                                                              ⁢                                                                  f                        Tx                                                  (                          l                          )                                                                    ⁡                                              (                                                  k                          -                          n                                                )                                                                                                                  +                                          n                j                            ⁡                              (                k                )                                                    ,                            (        5        )            where hjl corresponds to the j-th row and l-th column element of the matrix H with CSI (channel state information), and nl(k) is the j-th entry of the additive noise vector nk. In order to retrieve the signal from the first transmitting antenna, the received signals are filtered using filter 104 with impulse response gRx(l)(n) (=1/fTx(l)(n)), which is an inverse of fTx(l)(n). Thus, the signal received from each antenna after removing the tag associated with the first signal may be expressed as:
                                                        u                              1                ,                j                                      ⁡                          (              k              )                                =                                    ∑                              m                =                0                                                              N                  R                                -                1                                      ⁢                                                            v                  j                                ⁡                                  (                                      k                    -                    m                                    )                                            ⁢                                                g                  Rx                                      (                    1                    )                                                  ⁡                                  (                  m                  )                                                                    ,                            (        6        )            where NR is the length of the impulse response gRx(l)(n). In general, gRx(l)(n) is designed such that
                                                        ∑                              n                =                0                                                              N                  R                                -                1                                      ⁢                                                            f                  Tx                                      (                    l                    )                                                  ⁡                                  (                                      k                    -                    n                                    )                                            ⁢                                                g                  Rx                                      (                    l                    )                                                  ⁡                                  (                  n                  )                                                              ≈                      δ            ⁡                          (                              k                -                D                            )                                      ,                            (        7        )            where, δ(n) is the Kronecker delta represented by
                              δ          ⁡                      (            n            )                          =                  {                                                                      1                  ,                                                                                                  if                    ⁢                                                                                  ⁢                    n                                    =                  0                                                                                                      0                  ,                                                                              otherwise                  ,                                                                                        (        8        )            and D is a delay that arises from the non-minimum phase of the TT-Filter (tagging filter). From Eq. (6), we can therefore say that as the order of the maximum-phase portion of filter fTx(l)(n) increases, the delay D will also increase.
In such an OFDM system, the unique tagging filter can be placed in position marked by <1> in FIG. 2, at the transmitter. FIG. 2 illustrates the configuration of an OFDM-MIMO transmitter. Assuming that N is an integer equal to or larger than 2, this transmitter sends data using N pieces of transmitting antennas #1 to #N. Hereinafter, a transmitting antenna with antenna number of j, that is, the j-th transmitting antenna is represented as “transmitting antenna #j.”
The transmitter includes: channel coder 111 for performing channel coding of a binary input signal; S/P convertor 112 for performing serial-to-parallel conversion for the channel-coded signal to map the data to each antenna; S/P convertors 113 for performing serial-to-parallel conversion for the data, for each output of S/P convertor 112, to map the data to subcarriers; IFFT units 114 which are provided for each S/P convertor 113 and perform inverse fast Fourier transform for the parallel data from the corresponding SIP convertor 113; P/S convertors 115 which are provided for each IFFT unit 114 and convert the parallel output of the corresponding IFFT unit 114 to a serial signal; CP adding units 116 which are provided for each P/S convertor 115 and add a cyclic prefix (CP) to the data; tagging filters 117 provided at the outputs of CP adding units 116 as the position <1> described above; and signal processors 118 provided at the outputs of tagging filters 117 to perform signal processing for transmission.
In this transmitter, S/P convertor 112 for performing the antenna mapping has N pieces of outputs. S/P convertor 113 for performing the subcarrier mapping, IFFT unit 114, P/S convertor 115, CP adding unit 116, tagging filter 117 and signal processor 118 for transmission are cascaded in this order and connected to each output of S/P convertor 112. The output of each signal processor 118 is supplied to the corresponding transmitting antenna of transmitting antennas #1 to #N.
Here, processing such as scrambling, CRC (cyclic redundancy code), FEC (forward error correction), interleaving and modulation is carried out in channel coder 111. Processing such as up-sampling, filtering by a band filter, predistortion, amplification is carried out in signal processor 118.
In such a transmitter, the position of the tagging filters is not limited to position <1>. It is possible to relocate the tagging filters at position <2>, i.e., position of outputs of IFFT units 114, or position <3>, i.e., position of outputs of S/P convertors 113, or position <4>, i.e., position of outputs of S/P converter 112.
At the receiver, inverse tagging filters will be designed such that they satisfy Eq. (7) and Eq. (8) described above. The position of these inverse tagging filters is shown in FIG. 3. FIG. 3 illustrates an example of configuration of a receiver in the OFDM-MIMO system.
The receiver shown in FIG. 3 is provided with N pieces of receiving antennas 103. The receiver includes: analog processors 1 which are provided for each receiving antenna 103 and perform amplification, modulation, analog-to-digital conversion and the like of the received signal; digital processors 122 which are provided at the outputs of analog processors 1 and perform processing including fast Fourier transform (FFT); M pieces of inverse tagging filters 123; MIMO signal processor 124 which is supplied with signals from the M pieces of inverse tagging filters 123 and performs MIMO signal processing; P/S convertors 125 which are provided for each of M sets of parallel outputs from the MIMO signal processor 124 and convert the corresponding parallel output to a serial signal; P/S convertor 126 which is supplied with the outputs of respective P/S convertors 125 in parallel and converts these outputs to serial data; and channel demodulator 127 provided at the output of P/S convertor 126 to perform channel decoding and demodulation. Each of N pieces of digital processors 122 includes M pieces of outputs which are inputted to any of M pieces of inverse tagging filters 123. Therefore, any inverse tagging filter 123 is supplied with the outputs from the M pieces of digital processors 122.
Each digital processor 122 performs processing such as OFDM frame synchronization, down-sampling, removal of the cyclic prefix, and fast Fourier transform (FFT). P/S convertor 125 may or may not be provided with a deinterleaving matrix. If P/S convertor 125 includes the deinterleaving matrix, P/S convertor 125 performs the interleaving based on the deinterleaving matrix. P/S convertor 126 performs demapping of the data from the antennas.
However, the following problems need to be solved in the MIMO system based on tagging filters:
(1) For explanation purposes, consider a case where second order allpass filters are used, and that all the poles of the filters are located at a radial distance of 0.5 from the center (origin) of a complex plane (z-plane). As illustrated in FIG. 4A, when there are two sources or antennas, the maximum separation (Ps) between the poles of these two sources will be “1”. As illustrated in FIG. 4B, when there are eight sources or antennas, the maximum separation between two poles will be |0.25 √{square root over (2)}(1+i)−0.5|=0.3827. Thus, as the number of sources increases, the similarity between two filters whose poles are adjacent to each other will also increase. Differentiating signals which have been tagged using filters, whose poles are located close to each other, will therefore become increasingly difficult. The increase in order will imply an increase in the number of multipliers and hence the system's complexity will be also increased, which increases the system's in cost.
(2) Similarly, as the order of an allpass filter increases, the delay associated with its inverse function will also increase. Such a long delay is not preferable for systems used in delay sensitive applications. Example of such a delay sensitive system includes equipment such as unmanned space crafts which are controlled remotely. The delay may also causes an annoyance during conversation by voice call, if it exceeds a certain threshold acceptable to humans.
It should be noted that the present inventor has proposed a system having a plurality of antennas or sources in which each source is filtered by a unique filter, in PCT publication WO2006/059767 [A1].
Japanese Patent Laid-open Application No. 2000-286821 (JP-2000-286821A) [A2] discloses an OFDM communication device which uses a small number of pilot symbols to compensate receiving distortion due to variation in multipath fading in a transmission path (channel) and is capable of reducing transmission path estimation error due to noise. This device uses a filter having an inverse characteristic for the characteristic of the transmission path, filters a signal corresponding to the channel state information and obtains the channel state information in the time domain by multiplying the pilot data by a signal after FFT.
Japanese Patent Laid-open Application No. 2002-111756 (JP-2002-111756A) [A3] discloses a radio communication system in which filters are provided within both the transmitter and receiver and which performs fading distortion compensation suitable for a propagation environment thereby reducing the bit error rate. In this communication system, a bandwidth of a signal is restricted before transmission by a filter having a bandwidth narrower than the maximum bandwidth of the input signal.
Japanese Patent Laid-open Application No. 2005-318117 (JP-2005-318117A) [A4] discloses a closed-loop type MIMO transmission system in which data transmission is carried out by selecting different coding rates and different modulation schemes among transmitting and receiving antennas and streams. In this system, a scrambler is used for mapping data to the different antennas.
Japanese Patent Laid-open Application No. H05-219488 (JP-5-219488A) [A5] discloses a system for transmitting and receiving a motion picture signal in which an interleaver and a scrambler are disposed in a transmitter and a deinterleaver and a descrambler are disposed in a receiver. In this system, the transmission signal is subjected to an interleaving process and a scrambling process before NRZI (non-return-to-zero inversion) conversion, which is regarded as one kind of modulation, is performed.
Japanese Patent Laid-open Application No. H11-215091 (JP-11-215091A) [A6] describes an OFDM signal transmission system in which a transmission data sequence is subjected to a scrambling process or an interleaving process at a transmitter before modulation and a signal after demodulation is subjected to a descrambling process or a deinterleaving process at a receiver.
In the following, references referred to in this description are enumerated:                [A1] WO2006/059767        [A2] JP-2000-286821A        [A3] JP-2002-111756A        [A4] JP-2005-318117A        [A5] JP-5-219488A        [A6] JP-11-215091A        [B1] G. J. Foschini and M. J. Gans, “On limits of wireless communications in a fading environment when using multiple antennas,” Wireless Personal Communications, Vol. 6, pp. 311-335, 1998.        [B2] A. Goldsmith, S. A. Jafar, N. Jindal and S. Vishwanath, “Capacity limits of MIMO channels,” IEEE Journal on Selected Areas in Communications, Vol. 21, No. 5, pp. 684-702, June 2003.        [B3] Y. Ogawa, K. Nishio, T. Nishimura and T. Ohgane, “A MIMO-OFDM system for high-speed transmission,” 2003 IEEE 58th Vehicular Technology Conference, Vol. 1, pp. 493497, 2003.        [B4] H. Yang, “A road to future broadband wireless access: MIMO-OFDM-based air interface,” IEEE Communications Magazine, Vol. 43, No. 1, pp. 53-60, January 2005.        [B5] A. Belouchrani and M. G. Amin, “Blind source separation based on time-frequency signal representations,” IEEE Transactions on Signal Processing, Vol. 46, No. 11, pp. 2888-2897, November 1998.        [B6] T. L. Marzefta, “BLAST training: Estimating channel characteristics for high capacity space-time wireless,” Proceedings of 37th Annual Allerton Conference on Communication, Control, and Computing, pp. 958-966, Sep. 22-24, 1999.        [B7] Y. Li and L. Yang, “Semi-blind MIMO channel identification based on error adjustment,” Proceedings of IEEE International Conference on Neural Networks and Signal Processing, Vol. 2, pp. 1429-1432, Dec. 14-17, 2003.        [B8] Y. Li and K. J. R. Liu, “Adaptive blind source separation and equalization for Multiple-Input/Multiple-Output systems,” IEEE Transactions on Information Theory, Vol. 44, No. 7, pp. 2864-2876, November 1998.        [B9] J. Okello and M. Ikekawa, “Tagging-filter based blind MIMO equalization without signal permutation,” IEEE Globecom 2005, Vol. 4, pp. 2145-2149, November-December 2005.        [B10] W. Zhuang and V. Huang, “Channel preceding using phase distortion for slowly fading channels,” Proceedings of Canadian Conference on Electrical and Computer Engineering, pp. 234-237, September 1994.        [B11] N. Maeda, Y. Kishiyama, H. Atarashi and M. Sawahashi, “Variable spreading factor-OFCDM with two dimensional spreading that prioritizes time domain spreading for forward link broadband wireless access,” IEICE Trans. Commun., Vol. E88-B, No. 2, pp. 487-498, February 2005.        [B12] D. N. Godard, “Self-recovering equalization and carrier tracking in two-dimensional data communication systems,” IEEE Transactions on Communications, Vol. 28, No. 11, pp. 1867-1875, November 1980.        [B13] K. N. Oh and Y. O. Chin, “Modified constant modulus algorithm: Blind equalization and carrier phase recovery algorithm,” Proc. IEEE ICC, Vol. 1, pp. 498-502, June 1995.        [B14] A. van der Veen, “Statistical performance analysis of the algebraic constant modulus algorithm,” IEEE Transactions on Signal Processing, Vol. 50, No. 12, pp. 3083-3097, December 2002.        [B15] P. A. Regalia, “A finite-interval constant modulus algorithm,” Acoustics, Speech, and Signal Processing, 2002. Proceedings. (ICASSP '02). IEEE International Conference on, Vol. 3, pp. III-2285-III-2288, 13-17, May 2002.        [B16] J. G. Proakis, Digital Communications, McGraw Hill, 4th Edition, pp. 840-852.        